by Vaclav Smil
Rossi’s first American commercial installation came in 1937 with German-built brass casting, but the need for rapid and massive expansion of wartime steelmaking favored the use of well-established methods, and so it wasn’t until 1947 that Rossi persuaded Alleghany Ludlum Steel to embrace the method. The first good-quality slabs (33 ×7.5 cm, and up to 10 m long) were made in May 1949 in the company’s Watervliet plant by an American-made (Koppers) casting machine, but this early attempt did not lead to a further commercial adoption. Junghans made his first successful continuous steel casting also in 1949 in his workshop, and in 1950 Mannesmann (the other owner of basic patent rights) acquired the rights and began to build the first German commercial line at Huckingen, and it began operation in 1952. Rossi expanded his promotion and licensing activities by establishing Continuous Metalcast Corporation and, in October 1954, Concast AG in Zurich. Its general manager, Swiss lawyer Heinrich Tanner, eventually wrote a detailed definitive history of the early era of this fundamental technical innovation (Tanner, 1998).
During the next three decades, Concast dominated the global expansion of continuous casting through two highly profitable arrangements. In return for licensing its key continuous casting patents the company first received, gratis, all information and patents arising from the operation of licensed plants. This arrangement was further strengthened in 1970 when Concast made a cooperative patent exchange agreement with Mannesmann, its principal competitor. Second, builders of casting machinery channeled their sales contracts solely through Concast, receiving in return worldwide marketing, and consulting services guaranteed to resolve any technical problems arising during the early phases of commercial operation. As a result, Concast had eventually controlled more than 60% of the global market for continuous casters, it was identified as a virtual monopoly, and in 1981 it had to be reorganized pursuant to antitrust rulings in the United States and Europe.
Diffusion and Improvements
Diffusion of continuous casting was only slightly different from the experience with BOFs: Japan raced ahead, and while the US steel industry was not such a laggard as it was with oxygen furnaces, it moved slower. Although the first Japanese continuous casting line began to work in 1955, most of the expansion took place during the 1970s when the continuous casting share rose from 6% to 60%, and by 1990 more than 90% of Japan’s steel was made that way (Okumura, 1994). The first continuous casting plants in the United States went into operation in 1962, at US Steel in South Chicago and Roanoke Steel in Virginia, followed by a fairly rapid adoption during the late 1960s (Hogan, 1971). As a result American continuous casting capacity rose from just 65,000 t in 1962 to 14.4 Mt by 1970, and the years between 1968 and 1977 saw the largest number (more than 1100) of new continuous casting patents, with half of them filed by American inventors, but the US adoption of the technique continued to lag: by 1975 continuous casting reached just 10% of US steel production, it rose above 20% by 1980, and it reached 66% by 1990 (Aylen, 2002; Burwell, 1990).
By the century’s end, the technique became universal, claiming nearly 87% of the global steel output, virtually all (96–97%) production in Japan, the EU, and the United States and 87% of the total in China, with only Russia (at 50%) and Ukraine (at a mere 20%) lagging far behind. The earliest models of continuous casting machines had simple vertical designs and hence they required tall structures or deep excavated pits. Designs capable of bending the hot metal strand (first used in 1956) reduced the assembly’s height by nearly a third; curved bow-type casters, commercially available since 1963 (and developed independently in Switzerland at von Moos Steelworks and in Germany by Mannesmann), brought further height reduction by at least 40%, and low-head and horizontal designs became common starting in the 1980s (Okumura, 1994).
The standard sequence of modern continuous casting is as follows (Luiten, 2001; Morita & Emi, 2003; Schneider, 2000; Schrewe, 1991; Tanner, 1998). Molten steel (from the oxygen or electric arc furnace) is poured into a ladle that is transported to the inlet of a continuous casting machine, raised onto a turret, and poured into a tundish, a vessel whose large volume ensures the continuing flow of hot metal through a submerged entry nozzle and into a water-cooled and vertical copper mold, where the solidification begins as an outer shell starts enveloping a liquid core. The mold oscillates in order to minimize friction and eliminate any sticking, shell tearing, and liquid steel breakouts that would stop the casting and require expensive cleanup and repairs. As soon as the metal’s edge zone solidifies (1–2 cm), the withdrawal unit draws the strand out and the rollers move the partially solidified strand at speeds matching the flow of new supply.
Casting speed depends on the profile and quality of the cast metal, and for standard slabs, billets, and blooms it may be as slow as 30 cm and as fast as 7.5 meters per minute, with the most common rates between 1 and 2.5 meters a minute. Primary cooling takes place in the mold zone; the secondary cooling is done by water sprays as a new slab is drawn forward and gets bent on support rolls without any deformation or cracking. The strand gets straightened, and at the caster’s end, between 10 and 40 m from the hot metal inlet, even its core becomes solid and the metal is cut by oxyacetylene torches and discharged to be either stored before further processing or hot-charged for final rolling (Fig. 5.4). The process can be configured to produce a variety of semifinished steel castings. Early designs would cast only square or round billets of limited dimensions. Later machines could handle much larger blooms and slabs and produce near-net-shape profiles.
Figure 5.4 Torches cutting continuously cast steel at the Novolipetsk Steel mill in Kaluga, Russia. Corbis.
The most rational step would be to move the continuously cast hot slabs directly to the rolling mill to make finished products, a sequence that would bring further energy savings and reduce the inventories of semifinished profiles, but established plant layouts do not usually allow such additions, and the high capital costs of retrofitting make it less appealing. More importantly, since the late 1980s it has also been possible to make thin slabs (5–6 cm thick, casting speed of 4–6 meters a minute, to be hot-rolled directly into strips), and a decade later also thin strips (1–5 mm, casting speed 15–120 meters a minute), that can be coiled at the end of the process. Development of thin slab casting started during the 1970s, and Germany’s Schloeman Siemag, a leading maker of casting machinery, had the first prototype in 1985. Nucor in Indiana operated the first thin slab caster in 1989, and during the 1990s other designs were introduced in Japan, Italy, Austria, and Sweden.
The greatest challenge was to master direct production of thin strips, an effort that involved a number of secretive, and expensive, development projects in all major steel-producing countries during the 1990s (Luiten, 2001). The process does not use any oscillating molds, as molten steel goes directly between two water-cooled drums, and as it cools about 1000 times faster than during slab casting, it gets compressed by rotating drums into strips just a few millimeters thick (Schneider, 2000; Takeuchi et al., 1994). In a typical bow-type caster used to produce steel strips, temperature falls from 1500 °C to 1150 °C by the time the metal leaves the withdrawal rollers, and rapid rates of cooling (with heat fluxes reaching as much as 20 MW/m2 and temperature differences of up to 300 °C/cm in the forming strip compared to just 1–5 MW/m2 in oscillating mold casters) allow high casting speeds of more than 100 meters a minute, but it is a challenge to avoid premature solidification at edges and to produce a uniform strip with the exit temperature of about 650 °C.
Nippon Steel and Mitsubishi Heavy Industries began to operate the first strip-casting line (1.33-m-wide and 2–5-mm-thick stainless strip) in secret in the fall of 1997 and announced its success a year later, when Australia’s Broken Hill Proprietary Company also revealed its line at Port Kembla in New South Wales capable of casting strips of 1.5–2.5 mm that were reduced by an inline rolling to 1.1 mm (Bagsarian, 1998). Cooperation of three European companies resulted in the Eurostrip (for austenitic stainless steel, 1.5–4.5 mm thick and
1.1–1.45 m wide, average casting speed 60–100 meters a minute, coil weight 30 t) launched at ThyssenKrupp’s Krefeld plant in December 1999 (Bagsarian, 2000).
Continuous casting is now a mature production process, with new casters built and some old ones reconfigured, and with incremental performance improvements aimed at increasing throughput with higher casting speeds and at reducing the incidence of costly breakouts when molten steel flows through a defective part of a solidifying shell and makes costly damage to the casting line that requires extensive repairs. In Japan, these efforts to raise the net working ratio of caster lines have lowered the average incidence of breakouts per caster from 5.5 per year in 1990 to 2 in the year 2000 and to less than 1 incident in 2011 (Yamaguchi, Nakashima, & Sawai, 2013).
Continuous casting offers six major advantages when compared to traditional ingot casting that is followed by primary rolling: the process is an order of magnitude faster (30–60 min compared to 1–2 days for the same mass of processed metal); its metal yield is significantly higher (as much as 99% compared to less than 90%); it saves 50–75% of energy; its labor productivity is also at least 50% and up to 75% higher; capital savings are commonly around 60% compared to the traditional setup, and elimination of expensive hot strip mill by direct strip casting can lower the overall capital expenditure by up to 90%; and it requires much less space (Okumura, 1994; Schneider, 2000).
Not surprisingly, this combination of advantages ensured the technique’s rapid worldwide adoption. And continuous casting provided the last link in the modern steelmaking sequence which has been most commonly installed in mini-mills. By 1950, the dominant sequence was from blast furnace to open hearth furnace to cast ingots processed after reheating in a rolling mill; by the year 2000 that sequence might, or might not, include blast furnace, its second step was either BOF or EAF, and the crowning achievement was continuous casting of products ranging from massive slabs to thin coiled strips.
The final step is to turn the cast steel into shapes that will leave the mill. The most important products from continuous casting of slabs are hot-rolled and cold-rolled sheets, coated sheets, and electrical sheets. Slabs put through a plate mill end up as plates or UOE pipes (Fig. 5.5). Billets are turned into H-shapes, sheet piles, reinforcing bars, wire rods, and seamless pipes, and blooms end up as rails and light and heavy sections (Cullen et al., 2012). Some steel products are heat treated: this is done in order to harden the metal (using natural gas to heat them to specific temperature before rapidly cooling them by quenching into oil, water, or brine); to temper it in order to reduce brittleness (letting it cool after heating); and to anneal it in order to make it more ductile (keeping it at a specified temperature for a prescribed period, then cooling it slowly).
Figure 5.5 Slab casting at JFE’s Fukuyama Works (Hiroshima province). Reproduced by permission from JFE Steel.
Detailed, and reliable, Japanese statistics allow us to follow the partitioning of crude steel as well as all major finished products (JISF, 2015). Data for 2013 show all but 0.3% of pig iron (used by foundries and ferroalloy production) going for steelmaking. Nearly 80% of all crude steel (about 79% of it coming from BOFs and the rest from EAFs) is sold as ordinary steel; the rest are specialty products. About 89% of steel is hot-rolled, with heavy plates, bars, and shapes being the three most common products made from ordinary steel, and strips, bars, and wire rods dominating hot rolling of specialty steel. The breakdown of all finished steel products is as follows (all shares are rounded): hot-rolled strips 18%, galvanized sheets 13%, bars and plates each 11%, shapes and cold-rolled sheets each 7%, and wire rods 2%.
Chapter 6
Materials in Modern Iron and Steel Production
Ores, Coke, Fluxes, Scrap, and Other Inputs
Abstract
The next logical step—after tracing the history of iron and steel production, and after focusing on the role of technical advances in raising output and productivity, expanding the choice of final products and improving their quality—is to go beyond the confines of the plants where smelting, casting, oxygen blasting, rolling, and finishing take place, as well as looking at the production processes from the perspective of energy costs and environmental consequences. The latter two subjects will be the topics of the next chapter; in this one I will take a closer look at the key material inputs required to produce pig iron and steel.
Keywords
Iron ores; sintering and pelletization; fluxes; steel scrap; recycling; furnaces; material balances
The next logical step—after tracing the history of iron and steel production, and after focusing on the role of technical advances in raising output and productivity, expanding the choice of final products and improving their quality—is to go beyond the confines of the plants where smelting, casting, oxygen blasting, rolling, and finishing take place, as well as looking at the production processes from the perspective of energy costs and environmental consequences. The latter two subjects will be the topics of the next chapter; in this one I will take a closer look at the key material inputs required to produce pig iron and steel.
These inputs are dominated by the requirements of pig iron smelting, by massive (and increasingly intercontinental) deliveries of iron ores (now rarely charged as raw mine products but only after being subjected to specific beneficiation and agglomeration processes), coke, and fluxes (mostly limestone)—but they also include less massive, but no less essential, supplies of powdered coal, cooling water, and refractory materials. As already noted, the maximum daily output of large (>3000 m3) blast furnaces (BFs) is on the order of 13,000 t of pig iron, a flow that necessitates massive material inputs on an annual basis. Such furnaces require every year more than 4.5 Mt of hot air blast, and their total consumption of agglomerated ores (as sinter or pellets, now rarely as crushed ore), coke, injected coal, and fluxing materials (mostly incorporated in agglomerates) sums up to more than 12 Mt of raw materials, all to be charged, in an essentially punctiform way, through a furnace’s top. After appraising these inputs, I will present typical material balances for a number of current commercial BF practices.
Material requirements of basic oxygen furnaces (BOFs) are less complicated than the inputs into primary iron production: dominated by hot pig iron, they also include a significant amount of scrap metal, necessary fluxes, and, of course, a reliable supply of oxygen and cooling water. Modern steelmaking has become increasingly dependent on recycled metal, and before I will quantify typical material balances of electric arc furnaces (EAFs), I will present key information on the categories, generation rates, accumulated stocks, recycling rates, and international trade of steel scrap, as well as on the limitations of ferrous recycling.
Materials for BFS and BOFS
BF burden used to consist of crushed and sized iron ore and fluxing minerals (mostly limestone) charged with coke (to support the burden and to act as the reducing agent), all added via mechanical conveyors through a sealed top. Typical modern burden is a sintered or pelletized ferrous charge that incorporates flux materials, and reduced amounts of coke accompanied by direct injections of pulverized coal (the latter charge introduced through tuyères rather than through a sealed top). Detailed American statistics allow us to trace the reduction of specific raw material inputs since the beginning of the twentieth century (Gold et al., 1984; Kelly & Matos, 2014).
For iron ore (and/or its agglomerates), average charge per tonne of pig iron declined from about 2.1 t in 1900 to less than 2 t by 1925 and changed little during the next 25 years; by 1960 it was down to about 1.75 t, in 2000 it was just above 1.5 t, by 2010 it was just below 1.5 t, and in 2013 it averaged 1.35 t (WSA, 2015). Specific coke charging declined steadily from about 1.3 t per tonne of pig iron in 1900 to just below 1 tonne by 1932; little changed during the next two decades, but by 1960 the rate was about 820 kg/t, in 1970 it was reduced to less than 650 kg/t, by 1990 it was about 500 kg/t, and it averaged only about 400 kg/t in the year 2000 and 320 kg/t in 2015.
But, as already explain
ed in the previous chapter, this decline does not reflect the overall decrease in carbon needs as it has been accompanied by increased direct coal injection, whose average rose from less than 100 kg/t during the early 1990s to maxima of about 200 kg/t by 2015. Japanese data (Naito, Takeda, & Matsui, 2015; Takamatsu et al., 2012) provide another illustration of this shift. In 1945, Japanese BFs averaged more than 1500 kg of coke/t of pig iron, by 1950 coke-only smelting needed just over 900 kg of coke, and by 1975 the coke ratio was down to 450 kg/t but commonly used oil injection was adding about 100 kg/t. Following the two rounds of rapid oil price rises, all Japanese BFs became oil-less by 1982 and pulverized coal injection began to rise to reach an average of about 130 kg/t by the century’s end, in addition to 370 kg of coke, with a maximum monthly PCI rate of 266 kg/t and minimum coke use of less than 300 kg/t.
The US national mean for charging of fluxing minerals (limestone and dolomite) was around 380 kg/t of pig iron during the earliest decades of the twentieth century; subsequent slow rise brought it to as much as 480 kg/t by 1948, and then the rates declined to 330 kg/t in 1960, and during the last decade the rates were only around 275 kg/t. But the fundamental difference is not in charged rates but in the mode of typical use: with the near-universal adoption of agglomerated (beneficiated) ores, calcined limestone and dolomite are not charged directly as raw materials but overwhelmingly as self-fluxing sintered or pelletized ores.